CC3D and Revo Performance
« on: August 13, 2017, 11:02:53 pm »
Hello everone.

I scratchbuilt an acro quadcopter. You can see it here: https://forum.librepilot.org/index.php?topic=3721.0.
During the time I developed my piloting skills and I am pushing the quadcopter to the limits.
Will be a switch from F1 CC3D clone to F4 Revo flight controller an improvement for the flight characteristics?
« Last Edit: August 14, 2017, 11:32:39 pm by rad2014 »

Re: CC3D and Revo Performance
« Reply #1 on: August 14, 2017, 04:45:58 pm »
People would like to believe a different flight controller would make  there crafts fly better. It's all in  the tune how well they fly. The f4 board will comput faster  and have more room for more options. Like auto tune  that will be able to help you tune the quad better and might in turn help the flying characteristics. I would recommend you reading Everything you can on librapilot in the wiki section. The reason is so that you know what is available to you ,  When the going gets tough.

Re: CC3D and Revo Performance
« Reply #2 on: August 14, 2017, 11:46:36 pm »
Thanks for your advice. I am using Open and Libre Pilot since 2014 and during the time I have read most of wiki documenation.  I am familiar with differences between F1 and F4 CPUs. Do you (or anyone else) have  personal experience with CC3D replacement by Revo?

Re: CC3D and Revo Performance
« Reply #3 on: August 15, 2017, 03:42:46 pm »
I will soon, but as of right now I have only had experienced the joys of a cc3d.

Mateusz

  • *
  • 749
Re: CC3D and Revo Performance
« Reply #4 on: August 15, 2017, 03:59:26 pm »
I have used CC3D (F1), Revo and Sparky2 (F4) and SPRacing F3 and PicoBLX (F3) with "next" branch from git.
F3 and F4 have slightly more SRAM which means bigger programs can run in memory and therefore more advanced features can be enabled, example is Autotune.
Nevertheless, CC3D still have plenty of compute power and it can run pretty fast, Autotune sometimes fails, so EasyTune might be more robust option. Like Wayne mentioned, it's all about tuning the quad. This is actually what gives people edge in competitions. Not latest gyro, not super mega another shot protocol. You might want to enable active braking on ESCs though, might not be default. If you are really good pilot or consider newer hardware, the BLHeli_S ESCs (with busybee2 chip) are the best price/quality option at the moment. With such big build like yours, you might want to make sure that props are balanced, screws tight, and you dont have too much vibrations. If it flies bad, you might consider increasing filtering gyro/accel taus, but don't overdo it, and leave defaults if increasing them does not improve anything.

I would say CC3D is still a very good hardware, especially for rate/rattitude flying.

Re: CC3D and Revo Performance
« Reply #5 on: August 15, 2017, 08:28:50 pm »
Mateusz, I appreciate your answer very much.  To be honest you confirmed my presumption.
I have enabled active breaking on ESCs. I have softmounted the flight controller, put the gyro filtering to 0.000 and the quadcopter flies perfectly. During the time I fell in love to fly with Librepilot. I am flying with some Betaflight FC, too. BF behaves like a robot compared to Librepilot (no offence, just a comparison).

Mateusz

  • *
  • 749
Re: CC3D and Revo Performance
« Reply #6 on: August 16, 2017, 09:05:29 am »
Mateusz, I appreciate your answer very much.  To be honest you confirmed my presumption.
I have enabled active breaking on ESCs. I have softmounted the flight controller, put the gyro filtering to 0.000 and the quadcopter flies perfectly. During the time I fell in love to fly with Librepilot. I am flying with some Betaflight FC, too. BF behaves like a robot compared to Librepilot (no offence, just a comparison).

I am not religious about any particular flight-firmware, I would use whatever works for me. Having said, I tried and tuned Betafight (BF) on 3 builds ranging from 280mm to 80mm, so not judging without knowing anything. These quads are my father's so I kept BF on them for simplicity (that works ok for him) and on my own builds I use LibrePilot (LP) that works for me. Perhaps, I am not really good pilot to differentiate two tuned quads, but I am sure it would help LP developers a lot, if one could define "behaves like robot". It's like reporting an issue on JIRA, first you need to be able to define the problem, provide test case that can reproduce the issue, then developers can pick it up and work on it. They can never fix issue, if they can't reproduce it themselves . Maybe there is already a solution, but just different option has to be enabled and since these are different software it's hard to say what to change to get desired behaviour.
Each software has pros and cons, some features LP has, which BF has implemented differently, or not implemented at all, and the same goes of course other way around, to name a few 3 banks only LP, more advanced Autotune LP, dynamic Notch-filtering BF, anti-gravity-gain (BF) but this one is IMO some hack to solve symptoms rather than proper solution, it works but is not elegant and maybe there is more elegant solution already (I guess enabling thrust PID scaling would do the same thing). Since you are flying already both firmwares, and I assume you have both builds tuned to best of your abilities, it would be of really huge value, if you could point what's expected. For instance, when I do this manoeuvre banking or something, with this settings quad behaves like this with LP, but with BF it behaves like that, what to do, to make it behave like this (even with different hardware, maybe it will give idea what's different). As long as it is something better than "feels good" I am sure devs will follow it up :) Unfortunately, some blogs I read compared different quads, without saying what's wrong and showing one quad well tuned PIDs, other compared in that particular video I watched was LP quad using different hardware and totally way off PIDs shaking like crazy :) There goes conclusion, it  really had something like fair comparison in title but in description said it's stock PIDs as well...  ;D

Sorry for long reply!